:

ELIZA'S LEGACY: WHY PEOPLE CONFIDE IN CHATBOTS

AI DESK1 MIN READ
TUE, JUL 14, 2026

■ AI-SUMMARIZED FROM 2 SOURCES ▸ TIMELINE

Sixty years after MIT professor Joseph Weizenbaum created ELIZA, the first chatbot, people continue sharing secrets with AI systems like ChatGPT. The pattern established by early human-computer interactions now shapes how users engage with modern language models.

Weizenbaum's ELIZA, developed in the 1960s, demonstrated an unexpected phenomenon: users disclosed personal information to a machine they knew was not human. This early precedent explains current behavior with ChatGPT and similar systems. The chatbot's return to WhatsApp in Europe underscores growing AI integration in messaging platforms. OpenAI re-enabled ChatGPT access following EU pressure on Meta to open WhatsApp to competing AI services. The feature now operates across the European Economic Area, including the 27 EU member states plus Iceland, Liechtenstein, and Norway. This regulatory intervention reflects broader EU efforts to ensure interoperability and prevent monopolistic control of AI-powered services. The availability of rival AI options on WhatsApp represents a shift in how users access conversational AI within their existing communication channels. Weizenbaum's original work highlighted a psychological dynamic that persists: users treat responsive systems as confidants, regardless of technical limitations or privacy implications.

■ SOURCES

WiredThe Decoder

■ SUMMARY WRITTEN BY AI FROM THE LINKS ABOVE

■ MORE FROM THE AI DESK

Bill Gates has called for certain sectors to reserve jobs exclusively for humans as artificial intelligence advances. The Microsoft co-founder warned that governments remain unprepared for the technology's economic impact.

JUST NOWAI Desk

Pet owners, rescue agencies, and wildlife groups are demanding new protections as AI-generated images make it increasingly difficult to distinguish real animals from synthetic ones online.

1H AGOAI Desk

IBM has released the Granite 4.2 family of open-weight language models in three sizes under the Apache 2.0 license. The models feature built-in agentic capabilities and support context windows up to 512,000 tokens.

1H AGOIndustry Desk

A nonexistent US research organization, funded by Israel, published over 560,000 words across 124 reports in nine days using AI-optimized platforms designed to influence chatbot outputs.

1H AGOAI Desk

■ SUBSCRIBE TO THE DAILY BRIEF

ONE EMAIL, 5 STORIES, 06:00 UTC. UNSUBSCRIBE ANYTIME.